(FORTUNE Magazine) – A year ago, entrepreneur Dale Sundby was flying. PowerAgent, the company he'd formed to revolutionize advertising on the Internet, had just raised $13 million in its second round of venture capital financing--virtually overnight--to add to the $3 million it had raised 12 months earlier. From his spacious, glass-and-marble-furnished office suite high atop San Diego, Sundby envisioned PowerAgent as nothing less than the next Yahoo. And for a few months in early 1997, that idea didn't seem so crazy. Yahoo had gone public in early 1996 with a business model based predominantly upon selling advertising on the Web, and was already worth more than $1 billion. PowerAgent was going to sell more advertising than Yahoo--by creating cool, useful Internet ads tailored for specific people. If you were looking to buy a new stereo, for instance, PowerAgent would find you deals and promotions and then send them to you in splashy new ways. Sundby believed these personalized ads would transform online advertising from the mere sliver of the American advertising pie that it is today into the multibillion-dollar business that the prophets of cyberspace have always promised it would be. After conceiving the idea in 1994, Sundby had quit his job as CEO of an old-line San Diego law firm (earlier he'd done a ten-year stint at IBM), and sold his house for seed money.

He'd made believers out of a lot of people in the first two years of PowerAgent's existence. Dozens of experienced professionals abandoned comfortable, high-level jobs at places like Intuit and IDG, the computer-magazine publisher, to join Sundby and help fulfill his dream. PowerAgent's board of directors included such personages as former McKinsey honcho Fred Gluck and Arch McGill, the ex-president of American Bell. The list of investors included venture capital firms Information Technology Ventures (ITV), Draper Fisher Jurvetson, and St. Paul Venture Capital; Peregrine Investment Holdings, the Hong Kong-based and now defunct investment bank; Houston billionaire Fayez Sarofim; and, especially, Electronic Data Systems--PowerAgent's largest investor. John Fisher of Draper Fisher--a man with 13 years in the company-creation business--publicly proclaimed that PowerAgent had "one of the most powerful business models I've ever seen."

But in just ten tumultuous months last year, PowerAgent turned from a beautiful dream into a troubled, desperate mess. Mere weeks before Sundby's Big Idea was scheduled to be unleashed onto the Internet--it died. Sundby had blown through $22 million in seed money, venture capital, and loans and was unable to raise any more. He had to fire all but half a dozen of the 60 employees who had bought into his vision. Suddenly, PowerAgent wasn't the Next Big Thing, or even the next midsized thing. It was just a very big and very public object lesson for would-be Bill Gateses and Steve Cases: Sometimes even a great idea, an elegant business plan, smart people, and lots of money can't change the world.

Today Sundby, 45, lives on Silicon Valley's dark side--a place where there are no happy stories about secretaries becoming millionaires and where for every Yahoo there's a heap of smoldering, costly failures like PowerAgent. Each month Sundby digs deeper into his own pocket to pay for considerably less opulent offices. There he spends 20-hour days waging bitter war against EDS, which he believes is the evil corporation that killed PowerAgent. Sundby filed a "megalawsuit" on Feb. 25 in U.S. District Court in San Francisco. In it, PowerAgent alleges that EDS, while pretending to be its corporate partner, in fact conspired to drive the company into bankruptcy for purposes of acquiring its technology and other assets at fire-sale prices, so that it could incorporate them into EDS' recently formed Internet and electronic-services division. EDS did this, the suit alleges, by enthusiastically promising an equity investment of $10 million when it knew PowerAgent was running out of money, then reneging at the last minute. Had it not been for EDS' predation, Sundby thinks PowerAgent would be alive and well today and ready to file for an IPO.

The damages that Sundby is seeking represent the difference between the scant $1 million he estimates his brainchild is worth today and the amount PowerAgent would have been worth in the public market by the end of this year. That figure is $3.5 billion (yes, that's a "b"). Taking up the fight with Sundby is David Boies of Boies & Schiller, a noted litigator who is one of the leading lights in antitrust law and who is also currently working for the Justice Department on the Microsoft antitrust investigation. Says Boies: "I was very impressed with Dale and the company he built. It's a very impressive story, and he'll be an impressive witness." The formal charges against EDS, outlined in a 36-page complaint that reads, in places, like a novel, include fraud and breach of fiduciary duty. "I have 25 years of my life on the line," says Sundby. "I don't need to get whacked by EDS because they decide they're going to misbehave. But I will take them down with me." EDS has refused to comment about PowerAgent, citing company policy not to discuss matters under litigation.

PowerAgent is just one example, and probably recent history's most dramatic, of Silicon Valley's post-garage stage, where an orgy of easy money means that no startup actually ever has to live in a garage. Thanks to record levels of venture capital, tech companies are raising--and spending--more today than they have in the history of the Valley. Entrepreneurs who once might have had to scramble for money are now more likely to lose sleep worrying about raising too much. According to the Price Waterhouse/San Jose Mercury News venture capital survey, the average size of a round of venture funding for Silicon Valley companies has increased 50% in just the past four years, from $3.6 million in 1993 to $5.3 million in 1997. "A lot of startups are living on a diet of pate," observes Mike Moritz, a partner at Sequoia Capital. "They don't know the difference between how much money they need and how much money they want."

Few companies were as bewitched by the siren call of easy money as PowerAgent. After raising the $13 million in December 1996, the company started burning through it at a rate of $2 million to $3 million a month, with the blithe expectation that more would be on the way. In addition to its San Diego digs, PowerAgent opened offices in Menlo Park, Chicago, Los Angeles, and Seattle. It also commissioned two promotional videos that cost a total of $250,000 to produce, and employed seven high-level executives making six-figure salaries. Sundby regularly checked into the Sofitel, one of Silicon Valley's toniest hotels, when he was in town--even though techie etiquette views that as way extravagant for a company still spending other people's money. "It looked like GM, not a startup," recalls Mark Dubovoy, one of PowerAgent's directors and a partner at venture capital firm ITV. Money was flying out the door so fast that by July 1997--when EDS made the $10 million investment promise it later reneged on--PowerAgent was scraping bottom and knew it would need at least another $30 million before it could even contemplate generating revenue. In other words, even if Sundby is right about nefarious misdeeds by EDS, PowerAgent couldn't have been pushed over the cliff if the company hadn't led itself to the edge.

PowerAgent's investors knew they were making an expensive, high-risk bet. Sundby was leading PowerAgent in an all-out sprint to the finish. Rather than a piecemeal rollout of the service, PowerAgent planned to launch three separate ad-delivery products in June 1997--ads delivered to your E-mail, displayed as banners on your desktop, and flashed from your Web browser during the wait for pages to download.

Sundby had chosen to have PowerAgent's software developed by companies in Minneapolis and San Diego, an approach that is at least twice as expensive as building a team of engineers in-house but potentially much faster. And rather than building user registrations slowly through word of mouth, PowerAgent was going to amass a database of one million consumers by the end of 1998 by spending $10 million on a national advertising campaign and other promotions.

Investors might not have counted on marbled entryways or luxury hotels, but they supported, at least initially, Sundby's contention that if PowerAgent got to market quickly with a large, sophisticated product, then the company could claim a major piece of the Internet advertising business, generate $40 million to $60 million in revenues in its first year, and do an IPO in the second quarter of 1998. "I felt that we were chasing a very large opportunity," says one investor. "The business model, even with the most conservative projections, led to very large revenues and high margins in a relatively short period." Says another: "Sometimes you get people like Dale who succeed, and they have names like Bill Gates and Scott McNealy."

The first hints that Dale Sundby might not be the next Scott McNealy appeared as early as April. The company had hired a crackerjack sales team that was well connected with advertising agencies around the country. There were glossy promotional packets ready to be doled out. The only thing PowerAgent didn't have was a finished product. The software had been in development for more than a year, but there were problems. Big problems. There was no way, for instance, to track such things as how many times a person viewed an ad, and no convenient way to bill advertisers for extended campaigns--things large advertisers would deem essential when buying Internet ads.

"The product development had been run by people who never bought media before," explains Victoria Heineman Pieper, a former PowerAgent product manager, referring to Sundby's lack of expertise in advertising and the absence of any product-marketing people at the company prior to early 1997. When product managers with marketing backgrounds were hired, they told Sundby the software would have to be significantly reworked. Says one: "I took one look at the product right after getting there and said, 'You've got to be kidding. We can't launch anything that looks like this. This is awful. Awful.' " Sundby insists that the product simply needed minor revisions, but former employees think Sundby's denial was part of the trouble. "Dale had a certain arrogance that we were going to force ad agencies to do it our way because we were going to be the game in town," says a former PowerAgent engineer.

So PowerAgent's June product launch was rolled back to October, creating the need for an additional four months' worth of capital. Sundby didn't think this would be a problem. "There were no surprises," he says. "All we did is create the need for a larger amount of money to be raised." Much larger, actually. Instead of $10 million for the planned third round of venture capital financing, PowerAgent was going to need $30 million. It held extended conversations about potentially large investments with Softbank, the Japanese publishing and software conglomerate, and US West, but ultimately both companies decided not to do a deal.

By July the need for capital was acute: PowerAgent would be unable to make its payroll at the end of the month. Fisher and Dubovoy, who had both initially supported Sundby's plans for building a big company fast, now argued for slamming the brakes on spending. They wanted to scale down the company to a third of its employees, bring the software development in-house, attract a few thousand users, trickle along until the concept had been proved, and then ramp up again.

It was then that EDS--like the lifeguards on Baywatch dashing into the surf--came to the rescue. At a PowerAgent board meeting in late July, Barry Sullivan, EDS' vice president of Internet and electronic services, informed his fellow directors that EDS would make an additional investment of at least $10 million. Once that offer was on the table, other shareholders were willing to put up more as well. Sarofim would do an additional $6 million. St. Paul was in for another $3 million to $5 million. All together, it looked as if $20 million was on the way. "The board immediately started asking about approval process and what were the risks, but Barry assured us that it would go through smoothly and quickly," recalls Sundby. "He said that in 25 years at EDS, he had never been turned down for a funding project he'd sponsored."

Certain that it had the money from EDS to lead the new round of financing, PowerAgent continued spending aggressively to complete its software, and Sundby went ahead and formally unveiled PowerAgent to the public in early August. Following Web company protocol, this included a so-called cyber-elite party at a bar in San Francisco's "Multimedia Gulch"--the techie neighborhood just south of downtown. PowerAgent was the subject of dozens of articles in trade magazines and local newspapers and generally harvested the hype expected of all self-respecting Internet startups these days.

But as the schmoozefest advanced, EDS was backpedaling. In mid-August, Sullivan informed Sundby that EDS couldn't approve the investment until its September board meeting because he was going on vacation. For PowerAgent this was troubling news, since at the rate the company was spending, it would be millions of dollars in the hole by late September. Aghast, Sundby and his seven-person management team hopped a plane to EDS headquarters in Plano, Texas, for a meeting with Sullivan, who assured them that EDS was still excited about PowerAgent and intent on making the key $10 million investment.

Sundby was sufficiently nervous that he called an emergency board meeting at the offices of Venture Law Group, PowerAgent's attorney, on Sept. 15. There, Sullivan announced a sobering change of plans. EDS would invest only $5 million, and as a condition of the investment, EDS would take over the task of developing PowerAgent's software. Sundby says he was still suspicious but relieved he at least could tell employees the company was still alive.

Four days later Dave Williams, an EDS vice president who oversees EDS' venture capital investments, called with yet another, devastating, change of plans: EDS wasn't going to make an equity investment after all. Instead it would lend PowerAgent the $5 million. With the loan came a litany of terms so onerous that not even a company as desperate for money as PowerAgent could agree to them: The loan would be secured by exclusive technology and marketing rights to PowerAgent's products. PowerAgent would be in default if it failed to meet certain financial benchmarks each month. Also, EDS would get 10% of PowerAgent's revenue, the right to appoint a senior PowerAgent executive, and free warrants for PowerAgent's stock. After seven hours of arguments and accusations at another emergency board meeting on Sept. 24, EDS' terms were rejected, and PowerAgent's sad fate was sealed.

No one will ever know what could have become of PowerAgent if EDS had invested the $10 million. PowerAgent was about to sign a deal to be on Netscape's NetCenter page, which would have given it instant visibility on the Internet. Everyone has someone to blame. Sundby blames EDS. A number of former employees blame Sundby. Some board members blame one another. The only thing that just about everyone agrees on is that PowerAgent's concept is still viable: Someday someone will figure out how to make hundreds of millions of dollars by selling highly targeted advertising on the Internet. It just won't be Dale Sundby. Says Dubovoy: "I put this in the 'crying shame' category. It really could have worked."